Track genomics, neuroscience and molecular biology.
Search papers, labs, and topics across Lattice.
The authors upgrade Molecule-Morphology Contrastive Pretraining (MoCoP v2) by substituting hand-crafted CellProfiler features with deep-learning-extracted cell image representations to align small molecules with phenotypic perturbations. Bridging molecular graphs with rich, learned morphology embeddings unlocks morphology retrieval that scales log-linearly with training dataset size. MoCoP v2 substantially improves downstream QSAR linear probing and fine-tuning, beating multi-modal baselines trained on both morphology and transcriptomics on toxicity benchmarks.
Deep-learning-extracted cellular imaging features enable small-molecule representations that scale log-linearly with data volume and outperform models incorporating gene expression on downstream toxicity tasks.
Recent advancements in image-based profiling techniques have enabled the collection of high-volume cell morphology data, allowing new molecular embedding models to learn from the experimental phenotypic perturbations of a molecule in a cell. Previously, we developed Molecule-Morphology Contrastive Pretraining (MoCoP), a strategy for aligning small molecule embeddings to morphology fingerprints extracted through CellProfiler. The resulting molecular representation showed transferable performance for quantitative structure--activity relationship (QSAR) prediction tasks. Here, we extend the method by using a deep-learning-based cell image encoding pipeline to extract more feature-rich morphology profiles and align them to the molecular embeddings through contrastive learning. The new embeddings encode more accurate information on how molecules perturb cell morphology and enable improvements for QSAR predictions through either fixed-embedding linear probes or fully flexible fine-tuning. Morphology retrieval performance scales log-linearly with training data size, suggesting continued improvements as larger datasets become available. The improved MoCoP v2 also achieves superior performance on toxicity prediction and competitive results on ADME and activity benchmarks, when compared with existing molecular embedding models that use both cell morphology and transcriptomic data during training.